AMERICAN JOURNAL OF PHYSICAL ANTHROPOLOGY 56:297-304 (1981)

Cranial Capacity Estimates for Olduvai Hominid 7

MILFORD H. WOLPOFF Department of Anthropology, Uniuersity of Michigan, Ann Arbor, Michigan 48109

KEY WORDS Cranial capacity, Australopithecine, OH 7, Homo habilis

ABSTRACT Estimation of cranial capacity for Olduvai Hominid (OH)7 is de- termined from external parietal dimensions using multiple regressions calculated from an australopithecine grade sample. Capacity estimates for OH 7 (580-600 cc) are much lower than usually claimed. While differences in reconstruction may account for the varying estimates, a regression based only on undistorted and un- reconstructed values, as well as a direct comparison of dimensions with other Homo habilis specimens, supports the smaller capacity determination.

It would be an understatement to say that method, see Tobias, 1971).Holloway’s most re- times have changed since the discovery and cent approach, using multiple regression tech- initial interpretation (Leakey, 1961) of the ju- niques based on measurements taken from venile hominid specimen from Olduvai gorge, other endocasts appears to be more straight- OH 7. Still, some issues concerning the com- forward. Yet, his results still depend on the parative morphology of the specimen have yet accuracy of the reconstruction, as well as on to be settled. Of these, the question of the cra- the sample used to generate the regression. nial capacity of the vault represented by the With all due respect for Holloway’s experience two partial parietals has probably received the and appreciation of his substantial efforts, I greatest attention. wish to take issue with both of these points. Holloway’s most recent discussion of the OH 7 reconstruction (1980) is an important PROBLEMS IN THE RECONSTRUCTION contribution as well as a useful clarification of some longstanding issues. He has helped clar- ify the question of whether there was substan- I believe that there are some problems in the tial twisting in the reconstructed fit between reconstruction of the parietals that lead to un- the reconstructed parietals. However, other is- certainties in their alignment relative to each sues remain. It is clear, as Holloway states, other, and in their paracoronal curvature. The that the condition of the parietals, the amount correct determination of sagittal length and of bone preserved, and the accuracy of the re- biparietal breadth for the reconstructed pari- construction renders them anything but use- etal pair depends on both the reconstruction of less as a basis for volume estimations. Indeed, the coronal curvature and on the parietal I know of no worker who has claimed that alignment in the coronal plane. Moreover, they were unsuitable for this purpose. How- there is an obvious difference in paracoronal ever, suitability is one thing, and the appropri- curvature between the left and right bones. ateness of the estimation is quite another. It is These problems are further complicated by in this area that much of the controversy has the lack of any bone-to-bonecontact along the been focused. sagittal suture. Thus, it should be remember- The fact is that the parietals were crushed ed that the reconstruction of the parietal pair flat and have been reconstructed, and that be- that has been used in Holloway’s various fore Holloway’s most recent paper the volumet- analyses is based on a positioning of the two ric estimates based on this reconstruction parietals relative to each other that is deter- were determined from a biparietal partial en- mined by a set contact of bone with plaster docast procedure that has never been adequate- ly tested for accuracy (for details of the recon- struction and the biparietal partial endocast Received November 14, 1980; accepted July 24. 1981

0002-948318115603-0297$02.50 0 1981 ALAN R. LISS, INC. 298 M.H. WOLPOFF along the length of the sagittal suture. In the Here, there are two bone fragments that reach central portion of this suture, the contact is the midline at the lamboidal region. When ori- between plaster and plaster. ented along the midline as defined by the plas- ter sagittal suture reconstruction, the most Coronal curvature posterior aspect of the actual suture is posi- Each reconstructed parietal is made up of a tioned more superiorly than the most anterior number of bone fragments, clearly delineated aspect. The problem in this case is in the plas- by cracks that appear on both the internal and ter portion, but this is what defines the con- external surfaces. The positions of these frag- tact with the other side and the orientation of ments and their contacts are completely un- the combined parietals. ambiguous. However, I question some of the The effect of these problems on the recon- angles created between the individual frag- struction of the combined parietals can be out- ments at these contacts. For instance, on the lined by the contrasting biparietal breadth es- left bone, the paracoronal curvature of the timates that probably represent a maximum bone surface seems continuous over the infe- and minimum. Holloway uses the existing re- rior and central portions of the parietal wall. construction of the two parietals, mixing plas- However, at its most superior aspect, the two ter and bone, in which all sagittal suture on large fragments bearing the preserved (ante- one side is aligned by its contact with plaster rior) portion of the sagittal suture seem to be on the other. The maximum biparietal breadth set at an abrupt angle to the rest of the pari- for this reconstruction is approximately 115 etal. The sagittal suture is unduly elevated mm. I have used the criteria of approximate above the surface defined by the immediately verticality or only slight slope of the parietal contiguous bone fragments. As a result, if this sides, and alignment of the superior portion of parietal is aligned by positioning the inferior- the bones compensating for the angled frag- central portion vertically, the bone surface ments described above. In six determinations just below the sagittal suture-bearing frag- the maximum value for biparietal breadth I ments is oriented almost exactly horizontally, estimated is 102 mm. The poorly oriented su- but the two fragments with suture angle mar- perior left parietal fragment discussed above kedly upwards toward the midline (accentua- would seem to result in a keeled sagittal re- ting what I believe is their incorrect orienta- gion at this breadth. However, realignment of tion). This actually produces a concave groove this fragment to conform to the curvature of along a line roughly paralleling the sagittal the rest of the parietal would result in a flat- suture but between this suture and the tem- tened superior parietal region. poral line. On the other hand, if these suture- bearing fragments are used to orient the re- Sagittal alignment mainder of the parietal by positioning their The other problem to be noted in the present bone surface horizontally (asis required by the “set” reconstruction of the parietals is the bone-to-plaster contact of the present recon- alignment along the sagittal suture (i.e., along struction), then the central and inferior por- the plaster-to-bone contact). This is not a ques- tion of the bone has a very marked angulation tion of whether the sides are symmetric, or to the vertical and the maximum biparietal whether the parietals have been twisted rela- breadth, which is located along the inferior tive to each other (the propositions tested by border, is significantly expanded. I believe the Holloway, 1980), but rather whether one has present reconstruction is incorrect, both been translated relative to the other along the because of the angulation of the most superior sagittal plane. bone fragments (described above) and because The exact positioning of the parietals along of the angulation of the parietal sides to the ver- the sagittal plane is not simple or completely tical, which is unmatched in other early homi- obvious, since there are only a limited number nid sub-adults such as Taung (as suggested by of corresponding points preserved on the two the endocast form), SK 54 (Ron Clarke recon- bones. The posterior inferior corner and the struction), Omo 338y-6, and Modjokerto. most inferior portion of the lambdoidal suture The set of the most posterior (sagittal su- is preserved on both sides, as are many details ture-bearing) portion of the right side, relative of the meningeal artery impressions. When to the plaster reconstruction of the remaining the two fragments are aligned using the com- (anterior) sagittal suture-bearing region, parable landmarks on the posterior inferior shows that this side also contributed to what I corner, the “plaster bregma” on the right side regard as the inflated biparietal breadth. is at least 7 mm anterior to the real bregma OH 7 CRANIAL CAPACITY 299 preserved on the left. This indicates that there porary with a different australopithecine is a problem, and that the bregma-lambda di- grade species, Australopithecus boisei. More- mensions used by Holloway may be too long. over, that the range of cranial capacity varia- In an attempt to ascertain the correct align- tion in Homo habilis includes large-brained ment along the sagittal plane, I used three pre- specimens is clearly indicated by males such served points to orient the bones relative to as ER 1470. each other. These were the mastoid angle, the Thus the estimated capacity of this particu- posterior edge of the bevel for the temporal lar specimen is not critical with regard to the squama, and the base of the middle meningeal species characteristics of Homo habilis. If it artery impression. The parietals were reorien- has any importance beyond the accurate as- ted on three separate occasions, and in all sessment of the OH 7 characteristics, it is three cases the bregma-lambda distance was probably in reference to how cranial capacity less than along the existing plaster and bone can be estimated for incomplete specimens. It suture on the reconstruction. Interestingly, in is here that I differ from Holloway. his attempt to determine whether there is dis- As Holloway (1980) correctly assumed (see tortion of this reconstruction, Holloway con- footnote to Holloway’sTable 6),the contention cludes, “the most conspicuous region of distor- of a smaller capacity for OH 7 was based on the tion is on the left side, in the posterior region, external bony measurements. The two issues I roughly 2 cm from asterion (p. 271).” This am concerned with are whether external bony should alert one to a problem since the region measurements of the parietal can provide an is one of the few that can be directly compared estimation of cranial capacity for less than from side to side on the preserved bones. complete fossil hominid specimens, and how In sum, my reconstruction of sagittal length such an estimation Homo erectus and contem- me on the original specimens, assuring that 300 M.H. WOLPOFF the measuring points were defined the same sample of three in a single variable linear re- way and thus that the measurements are fully gression to predict cranial capacity. In this comparable. case the lambda-asterion arc gives a mean er- ror of only 28.3 mm. This arc was then combin- Choice of sample ed with each of the remaining five indepen- My choice of the sample used to determine dent variables resulting in five two-variable the multiple regression formula differs from multiple regressions, each of which exactly Holloway’s, who used varying base samples predicted the three cranial capacities used in with different proportions of hominids and its determination. pongids (three pongid species, Homo sapiens, These five regressions were then used to es- and an assemblage of fossil hominids span- timate the OH 7 cranial capacity. Omitting ning the last 3 million years). I disagree with the maximum and minimum estimates results this procedure because an estimation deter- in a sample of three estimates, averaging 574 mined by a multiple regression is most accu- cc (the standard deviation is 14.7). This esti- rately applied to the sample from which the re- mate is probably the most accurate that can gression was calculated (Pearson, 1926). In be determined for OH 7 from the Homo habilis this case there are only three Homo habilis sample alone. specimens with known capacities (OH 24, ER Of course, three specimens are a very small 1470, and 18131, all determined by Holloway sample. Although this is the most appropriate (1978).‘It should be mentioned that Holloway sample to use, there can be little confidence in expresses some uncertainty with respect to the results taken by themselves. A larger sam- the OH 24 capacity estimation of 590 cc. Yet, ple is clearly desirable. However, I believe the point he questions is whether this esti- that the choice of a larger sample should be mate may be too large (p. 383), especially in constrained by the limitations of what a re- view of some of its vault dimensions. Using gression does and where it is most validly ap- the maximum vault capacity reported for it in plied. Consequently, I contend that the next my study acts to weigh the case in favor of a best sample to use is determined by closeness large estimte for OH 7, since if the capacity es- of relationship, as dictated by grade and clade. timate for OH 24 really is too large, its use in a Because Homo habilis is an australopithecine regression would tend to cause overestimates grade species, other australopithecines are the of cranial capacity from vault dimensions. best sample from which to determine a predic- tive regression. However, of the recognized Five approaches australopithecine species, at least two clades The first approach in estimating a capacity are represented; Australopithecus africanus for OH 7 was to use the two complete Homo very likely is the species of the same clade that habilis specimens (ER 1813 and ER 1470) to is closest in grade to Homo habilis, and I be- generate nine different linear equations for lieve it is from this taxon that additional speci- cranial capacity as a function of each measure- mens for a wider based regression should be ment. Each of these equations was then used drawn. This provides five more specimens to estimate the OH 7 cranial capacity, based with cranial capacities determined by Hollo- on the measurement in question. The maxi- way (1978): STS 5, STS 71, TM 1511 (STSSO), mum and minimum determinations were omit- ted, and the average for the remaining seven capacity estimates is 578 cc (the standard de- ‘Holloway’s reconstructions are the most recent of numerous at- viation is 43.5). tempts for some of these specimens, and are widely used in the field. Whether or not they are correct for these more complete specimens The second approach was to include OH 24 (than OH 7). the fact that all of the comments addressed to this prob- in the Homo habilis sample. This resulted in a lem have proceeded from this assumption makes it necessary to accept smaLler number of usable measurements since the assumption of accuracy here. Otherwise, the discussion of tech- nique becomes confused with the question of whether the base sample all of the arcs could not be unambiguously de- is accurate. In this way, at least the starting point for each procedure termined for OH 24 (for instance, no arcs span- is the same. ning areas reconstructed in plaster were used). The mean error was determined as the average of the absolute dif- The cranial breadths used for this specimen ferences between the actual and calculated cranial capacities. This were all reconstructed symmetrically from the statistic was used instead of the standard error (or the adjusted standard error) so as not to unduly weigh outlying deviations in de- undistorted right side of the base. Since there termining the regression accuracy (the standard errors depend on the were more measurements (six)than specimens square of the difference). The coefficient of determination and the multiple correlation were not used because they provide a measure of (three), I first determined which single mea- the correlation between the actual and estimated values, and not a surement gave the lowest mean error2 for the measure of how closely these actually approximate each other. OH 7 CRANIAL CAPACITY 30 1 MLD 37, and the juvenile specimen Omo the sagittal suture, as discussed above. My es- 338y-6 [which, contra Rak and Howell (19781, I timates to these relations differ from HoUo- do not believe either represents a juvenile of way's, and I have used my own determina- A. boisei or differs from A. afn'canus in any tions in these regression calculations. significant way].3 This problem can be avoided if the contro- Using only the nine measurements that versial measurements are excluded from the could be determined for OH 74 (described regression attempts. Consequently, a fourth above), a multiple regression for predicting approach was employed in which only arc mea- cranial capacity was determined from these surements between landmarks connected by eight specimens. A stepwise approach was continuous bone surface on OH 7 were used to again employed, with each step chosen to min- determine a multiple regression from the eight imize the mean error of the predicted values specimens of Australopithecus afncanus and compared with the known values. The best fit- Homo habilis. In this analysis, all observable ting three-variable formula is given in Table effects of distortion and differing reconstruc- l.5The mean error for this formula is 6.4 and tions are removed since the reconstructed or the standard error is 11.7. This formula pre- questionable measurements are omitted. The dicts a capacity of about 600 cc for OH 7. The measurements used are the last five in the list best fitting four-variable formula (Table 2) has a mean error of 1.5 and a standard error of 4.6. The OH 7 prediction is about 605 cc. 'Rak and Howell cite "the marked prominence of the inferior tem- poral and the superior nuchal lines and the coarseness of the planum One could object to this third approach, ar- nuchale in an individual so young (p. 346)" as the primary support for guing that some of the measurements used re- placing this individual in Australopithecus boisei. However, the fact is that the specimen's age is unknown (for instance, it may have been present judgements made in reconstruction. a late adolescent with all but M3erupted), and in any event virtually In particular, the bregma-lambda arc depends nothing is known of the age-specific ontogeny of these features in any australopithecine group. A very well-developed masticatory appara- on the fit of the parietals over the sagittal su- tus and prominent lines of muscle attachment characterize some ture and, as mentioned above, there is no point South African Australopithecus africanus crania such as STS 71, and at which bone is present on both sides of the the elongated vault as seen from above more closely resembles this species than it does the globular form of the Australopithecus boisei suture. The maximum parietal breadth is in- vault. Holloway's recent analysis of the endocast (1981)also supports fluenced by the angulation of the parietals rel- the contention that this Omo juvenile probably is not Australopithe- ative to each other, and this in turn is affected cus boisei. 'As in the case of OH 24, only measurements that could be clearly by the reconstruction of vault curvature near defined and were not obviously altered by distortion were used for this comparative sample. % each case, the formula was only determined for those specimens in the comparative sample that had the full measurement set (these are specificially listed in Tables 1-4). Thus, no attempt was made to approximate missing data. TABLE 1. Best fitting three-uanable regression for the Homo habilis and Australopithecus africanus combined namnle.' Regression: Cranial capacity = 3.2* (biparietal breadth) + 9.3* (bregma-lambda arc) + 4.6* (lambda-asterion arc) - 921.9 mm TABLE 2. Best fitting four-uanable regression for the For the sample used in the regression Homo habilis and Australopithecus africanus combined samde.' Cranial capacity Regression: Cranial capacity = 3.5* (biparietal breadth) + Actual Predicted 12.3* (bregma-lambda arc) + 4.9* (lambda-asterion arc) OM0 338~-6 427 414 - 2.8* (bregmaparietal mastoid angle arc) - 949.3 mm. STS 71 428 436 For the sample used in the regression MLD 37 435 432 Cranial capacity STS 5 485 489 Actual Predicted ER 1813 509 520 OM0 338~-6 427 428 OH 24 590 587 STS 71 428 429 ER 1470 770 767 MLD 37 435 43 1 For OH 7, cranial capacity z 600 cc STS 5 485 486 'Measurements are in mm and capacity in cc. In this table, as ER 1813 509 511 well as in Tables 2 and 3, the equation coefficients and constant ER 1470 770 769 have been rounded to one decimal place. However, the actual capacity calculations were based on the unrounded values and thus For OH 7, cranial capacity a 605 cc will be slightly different from those based on the formulae presented. 'Measurements are in mm and capacity in cc. 302 M.H. WOLPOFF above. The best fitting four-variable regres- TABLE 3. Best fitting four-variable regression for the sion (Table 3) resulted in a cranial capacity es- Homo habilis and Australopithecus afiicanus combined sample, using only arc measures not reflecting timate of about 570 cc for OH 7, with a mean reconstruction or obvious distortion in OH 7l error of 10.5 and a standard error of 31.4 While this is not the most accurate of the de- Regression: Cranial capacity = 12.6* (lambda-asterion arc) + 13.4* (bregma-asterion arc) - 8.1* (bregma-krotaphion terminations in terms of the mean error, the arc) - 6.9* (lambda-parietalmastoid angle arc) - 663.9 mm fact that it is based only on measurements that were neither distorted nor affected by re- For the sample used in the regression construction in OH 7, and presumably would Cranial capacity be agreed upon by Holloway and me, suggests Actual Predicted that it might be the most realistic estimate. OM0 338y-6 427 420 Moreover, the fact that this estimate lies only STS 71 428 442 slightly below the two multiple regression es- MLD 37 435 446 timates discussed above indicates that the nar- STS 5 485 461 rower reconstruction may be the more accurate one since a broader vault would have resulted ER 1813 509 515 in even larger capacity estimates (see below). ER 1470 770 769 Finally, the expectation of a cranial capacity For OH 7,cranial capacity z 570 cc estimate for OH 7 between the values for OH IMeasurements are in mm and cranial in cc. 24 and ER 1813, and much smaller than the ER 1470 capacity, is supported by a comparision of these five undistorted arc measurements TABLE 4. The Homo habilis sample compared for cranial for the Homo habilis specimens (Table 4). capacity and for those parietal arc measures that do not reflect reconstruction or distortion in OH 7

DISCUSSION ER OH OH ER 1813 7 24 1470 In sum, five different regression attempts and a straightforward comparison provide sur- Cranial capacity 509 590’ 770 prisingly similar results for the estimation of Parietal arcs Bregma-krotaphion 76.5 80.0 88.0 OH 7 cranial capacity, clustering within about 25 cc (three within 8 cc). A value of approxi- Lambda-asterion 64.5 67.0 72.0 78.0 mately 580-600 cc would seem to be a reason- Lambda-PMA 76.8 85.0 87.0 88.0 able estimate, if not an overestimate, based on Bregma-asterion 113.0 121.0 132.0 these determinations, since the mean for the Bregma-PMA 106.5 109.0 120.0 five estimates is about 585 cc (with a standard ‘The initial capacity estimate (mean of eight determinations) was deviation of 16.0).If this exceeds the not “signi- 560 cc (Leakey et al., 1971). This higher, more recent, estimate is ficantly in excess of 500 cc” I have suggested used so as not to bias the OH 7 estimate downwards. before (Wolpoff and Brace, 1975), it falls even further from the 700-750 ml range given as the “most probable estimate” by Holloway (1980). the average ratio was chosen in a manner that It is reasonable to ask why these results dif- is similar to that used to pick the regression fer considerably from the estimates made sample for this study. Tobias (1971) reports using the biparietal partial endocast method that he used the average for four fossil homin- and from the estimates made from Holloway’s ids with reliable volumes: Taung, OH 5, and regressions. Indeed, why can Holloway report one specimen of Homo erectus from Indonesia that the OH 7 parietals fit over endocasts with (Trinil 2) and one from China (CKT 11).Thus, even larger volume? I believe that the answers the sample was also drawn from the most close- to these questions lie in two independent fac- ly related hominids, although covering a tors, the inflated size of the present recon- broader geographic range, a longer temporal struction and the choice of regression sample span, and a greater taxonomic diversity than I Holloway used in his estimation calculations. have done here. Yet there is sufficient similar- ity in method to question why the results are Partial endocasts different. In this case, I believe the answer lies The biparietal endocast method is itself a re- fully in the dimensions of the different OH 7 gression-type approach (although not actually reconstructions. a regression) since it is based on the average Using the dimensions taken on the current ratio of biparietal endocast volume to total en- OH 7 reconstruction (i.e., those used by Hol- docast volume. The sample used to determine loway) would elevate the capacity estimates OH 7 CRANIAL CAPACITY 303 for those regressions that employ sagittal arc plied to the sample from which it was calcula- and biparietal breadth, bringing them more ted, this surely must be the most appropriate in line with the biparietal endocast deter- regression set. The biparietal endocast values mination. Yet, apart from the argument that that have been accepted for almost two de- there are morphological reasons to question cades are based on a ratio average for an even the reconstruction that provides these larger smaller sample but covering a greater time- dimensions (see above), the fact is that other span, geographic range, and number of taxa regressions used here that do not use these (three taxa for four specimens), as well as mix- dimensions would not be altered (for instance ing specimens of two different evolutionary Table 3, or the majority of the determina- grades. The sample that Holloway used in- tions used to calculate the average given volves an extraordinarily broad range of time, under the discussion of the first and second space, and taxonomic diversity, and while the approaches above). resulting sample size is large, Holloway (1980) Thus, if the larger dimensions were used, admits: “The multiple regression analyses give there would be two different capacity estimate a wide range of values, depending on the sam- clusterings. I believe that this provides sup- ples chosen and their constituent members” [p. plementary evidence supporting the conten- 2731. Unfortunately, this is as it should be. tion that the sagittal and biparietal dimen- The question of which regression base sample sions of the current reconstruction are too is most appropriate cannot be resolved pre- large. Using the smaller dimensions that I cisely because none of them really are appro- have determined results in only a single clus- priate. Pearson (1926) questioned whether tering of capacity estimates. Moreover, the di- regression equations calculated in one human rect comparison of vault dimensions with the population could be validly applied to another, other more complete Homo habilis specimens and the biological diversity spanned by any of (Table 4)also indicates that the smaller rather Holloway’s samples is considerably larger. than the larger estimates (and consequently One final aspect of this problem is the criti- the dimensions from which they were deter- cism that, even if the sample I have used is va- mined) are more likely correct. The alternative lid, a number of the capacities that have been interpretation is that parietal dimensions can- used are questionable. For instance, Hollo- not be used to estimate cranial capacity be- way’s objections to his own OH 24 estimate cause they give inconsistent results (depend- have been noted, and one could similarly ques- ing on which set of dimensions are used). tion the STS 71 value because of the possibi- Should this be the case, it would follow that lity of distortion in the posterior region. none of the estimates that any of us have at- Similarly, Holloway and Tobias have arrived tempted have validity. at very different values for MLD 37. I believe this is a valid criticism, one which Different regression samples must be constantly borne in mind during any A second problem confounds the compari- discussion of the OH 7 capacity since every son with Holloway’s most recent capacity esti- one of these other questionable australopithe- mation. This is the choice of regression sam- cine specimens is more complete than OH 7. I ple. The only fully valid sample for generating have used only Holloway’s published capaci- a regression to estimate the OH 7 capacity is ties in my regressions for the sake of consis- the Homo habilis sample, and the determina- tency and because he had enough confidence tions from this sample provide the same small- in them to publish them. Clearly, however, the er regression estimates as the determinations OH 7 estimate can be regarded with no more from the larger sample I have chosen. None- certainty than the estimates for the other theless, the fact that the sample is small, and specimens used to determine the regression. the question of whether all of the specimens in Since these other specimens also form the ba- it should be considered Homo habilis (for sis of comparison of Homo habilis with AUS- instance, some workers regard ER 1813 as an tralopithecus afiicanus, the problem here is East African varient of Australopithecus afn- much broader than the single issue of how canus),could form the basis of valid criticism. large the OH 7 vault might have been. Yet, what are the alternatives? The larger sample I have chosen for comparison covers a SUMMARY wide timespan and a broad geographic range. In sum, the main differences in capacity es- Nonetheless, these are the specimens of simi- timates for OH 7 involve the reconstruction of lar grade that are most closely related to OH the parietal set and the choice of regression 7, and because a regression is most validly ap- samples used to calculate cranial capacity 304 M.H. WOLPOFF from the dimensions of the reconstruction. I uals, and the debt owed to them could never be have argued that the most appropriate basis overstated. I would also like to thank R.L. Hol- for such a regression is the hominid sample loway, who reviewed this paper several times most closely related to OH 7, and very similar and made substantive comments and sugges- results were obtained using two such samples tions that I believe have greatly improved it. (Homo hubilis alone and a mixture of all This work was supported by NSF grants those Homo habilis and Austrulopithecus BNS 76-82729 and GS-33035. ufricunus specimens with cranial capacities published by Holloway). I have further argued that, for a variety of reasons, the sagittal LITERATURE CITED length and biparietal breadth of the current Holloway, RL (1978)Problems of brain endocast interpreta- OH 7 reconstruction are too large. Using the tion and African hominid evolution. In Jolly, CJ (ed): smaller dimensions I ascertained during my Early Hominids of Africa. London: Duckworth. pp. 379-402. study of the original specimen, I have obtain- Holloway, RL (1980)The 0. H. 7 (Olduvai Gorge, Tanzania) ed regression estimates of the cranial capacity Hominid partial brain endocast revisited. Am. J. Phys. that are the same as the estimates based on re- Anthropol., 53:267-274. gressions that do not use the dimensions in Holloway, RL (1981) The endocast of the Omo L338y-6 ju- venile hominid: Gracile or robust Australopithecus? Am. question. For these reasons, I contend that J. Phys. Anthropol., 54:109-118. there is just cause for the claim that the OH 7 Leakey, LSB (1961) New finds at Olduvai Gorge. Nature capacity may have been considerably smaller 189:649-650. than the minimum estimate of 700 cc suggest- Leakey, MD, Clarke RJ, and Leakey LSB (1971) New ho- minid skull from Bed I. Olduvai Gorge, Tanzania. Nature, ed by Holloway. 232308-3 12. Pearson, K (1926)On the reconstruction of cranial capacity ACKNOWLEDGMENTS from external measurements. Man, 26:46-50. I am very grateful to M. Leakey, R.E.F. Rak, Y,and Howell, FC (1978) Cranium of a juvenile Aus- tralopithecus boisei from the lower Omo Basin, Ethiopia. Leakey, F.C. Howell, P.V. Tobias, A. Hughes, Am. J. Phys. Anthropol., 48:345-366. C.K. Brain, and E. Vrba for permission to ex- Tobias, PV (1971) The Brain in Hominid Evolution. New amine the original specimens in their care and York: Columbia University Press. for the hospitality and encouragement I receiv- Weidenrich, F (1943) The skull of Sinanthropus pekinensis; A comparative study on a primitive hominid skull. Palae- ed during my visits. Research in paleoanthro- ontologia Sinica D, No. 10. pology would be impossible without the active Wolpoff, MH, and Brace, CL (1975)Allometry andearly ho- and continuing cooperation of these individ- minids. Science, 189:61-63.